Capítulo IV. ¡Error! Marcador no definido.
4.5. Modelo de socialización de la propuesta
This section outlines the direction of effects for the particle placement variables that were statistically significant in both L1-L1 and L2-L2 data sets, before
152
introducing the ones that were significant in either data set. Figure 4.2 below illustrates the direction of direct object news value effect on the target type used. Figure 4.2: Direction of news value of the direct object effect on the use of verb- particle construction
We can see from Figure 4.2 that the L1-L1 and L2-L2 conversations display the same behaviour when the target’s direct object was mentioned in the five AS- Units preceding the target and when the direct object is entirely discourse new. There is an increase in the proportion of the VP PRT NP targets when the direct object is discourse new. However, it seems that given direct objects tend to favour the VP NP PRT particle placement pattern.
153
Figure 4.3: Direction of direct object syllable length effect on the use of verb-particle construction
Both data sets display the same kind of behaviour with regards to direct object syllable length predictor. Looking at Figure 4.4, we can see a greater preference for the VP PRT NP construction as the number of the target’s direct object syllables increases. Monosyllabic and disyllabic direct objects, however, are seen to favour the VP NP PRT particle placement pattern.
Figure 4.4: Direction of the direct object complexity effect on the use of verb-particle construction in the L1-L1 conversations
As was mentioned in section 4.4.2, complexity as a particle placement
predictor was included only in the L1-L1 GLM analysis. Figure 4.4 shows that direct objects that have embedded clauses are seen to prefer the VP PRT NP particle
154
placement target type. Direct objects that do not include embedded clauses, however, seem to favour the VP NP PRT particle placement pattern.
Figure 4.5 Direction of the direct object definiteness effect on the use of verb-particle construction in the L2-L2 conversations
The definiteness of the target’s direct object predictor was statistically significant only in the L2-L2 conversations, see Table 4.5 and Table 4.7. Figure 4.5 shows that direct objects that are definite tend to prefer the VP NP PRT pattern. The proportion of VP PRT NP pattern, however, is seen to increase when the target’s direct object is indefinite.
4.5.5 L1-L1 and L2-L2 particle placement interaction models
Table 4.9 shows the GLM results for the interaction of the priming related predictors and the speaker’s identity with the particle placement priming effect.
155
Table 4.9: The outcome of the interaction models in L1-L1 and L2-L2 particle placement priming
L1-L1 L2-L2
Chi2 test Chi2 test
Interactions Chi2 p AIC Chi2 p AIC
Previous type * prime- target distance
4.09 0.04* 230.38 1.29 0.26 234.57
Previous type * lemma match
1.02 0.31 235.55 8.85 0.003** 228.24
Previous type * shared particle
0.72 0.40 235.11 2.27 0.13 234.70 Previous type * D.O.
overlap
2.05 0.15 232.98 11.91 <0.001*** 225.46 Previous type * speaker
match
0.81 0.37 235.12 0.35 0.55 233.19
Note. L1-L1 = dyadic interaction between participants with English as their L1; L2-L2 = dyadic interaction between participants with English as their L2; D.O. = direct object; * = <0.05; ** = <0.01; *** = <0.001; Chi2 – test’s p-value tests an interaction between the prime and the other predictors included in the table, individually.
L1-L1 null model’s AIC = 233.10
L1-L1 independent priming predictor’s AIC = 232.64
L2-L2 null model’s AIC = 233.59
L2-L2 independent priming predictor’s AIC = 233.43
As we can see from Table 4.9, only in the L1-L1 conversations does the prime- target distance affect the size of the particle placement priming effect. In the L2-L2 conversations, we can see that both the shared prime-target main verb lemma and shared prime-target direct object overlap affect the size of the priming effect. These interactions will be further detailed in the following 4.5.5.1, 4.5.5.2 and 4.5.5.3 sections.
4.5.5.1 Prime-target pair distance
As we have seen in Table 4.5 and Table 4.7, prime-target distance was close to significance in the L1-L1 conversations (p = 0.091), but irrelevant to the type of the
156
particle placement used in the target in L2-L2 conversations (p = 0.299). Table 4.10 sh ows the descriptive statistics for all particle placement prime-target pairs distance in L 1-L1 and L2-L2 conversations.
Table 4.10: Descriptive statistics of all particle placement prime-target pairs distance in L1-L1 and L2-L2 conversations
Prime-target pairs Descriptive statistics L1-L1 distance L2-L2 distance Mean (SD) 16.08 (22.53) 22.50 (27.09) Median 6 13 Mode 1 1 Maximum 132 115 Minimum 0 0 Upper bound 19.04 26.41 Lower bound 13.13 18.60 Confidence level (95.0%) 2.95 3.91
Note. L1-L1 = dyadic interaction between participants with English as their L1; L2-L2 = dyadic interaction between participants with English as their L2; SD = standard deviation; Maximum = particle placement prime-target pair maximum distance in AS-Units; Minimum = particle placement prime-target pair minimum distance in AS-Units.
Table 4.10 shows that, compared to L2-L2 conversations, prime-target pairs were separated by about six AS-Units less on average in L1-L1 conversation. We can also infer from Table 4.10 that one AS-Unit gap separated most of the particle
157
Table 4.11: Descriptive statistics of the matched and unmatched prime-target pairs in the L1-L1 and L2-L2 conversations
L1-L1 prime-target pairs L2-L2 prime-target pairs Descriptive statistics Matched distance Unmatched distance Matched distance Unmatched distance Mean (SD) 13.65 (22.12) 21.04 (22.65) 21.81 (27.53) 23.63 (26.50) Median 5 13 11.5 14 Mode 1 1 1 1 Maximum 132 95 115 103 Minimum 0 0 0 1 Upper bound 17.16 26.57 26.87 29.91 Lower bound 10.14 15.52 16.75 17.36 Confidence level (95.0%) 3.51 5.52 5.06 6.27
Note. L1-L1 = dyadic interaction between participants with English as their L1; L2-L2 = dyadic interaction between participants with English as their L2; SD = standard deviation; Maximum = particle placement prime-target pair maximum distance in AS-Units; Minimum = particle placement prime-target pair minimum distance in AS-Units; Matched distance = prime-target pair distance where the prime and target are of the same verb-particle variant; Unmatched distance = prime-target pair distance where the verb-particle variant in the prime is different from the one in the target.
Table 4.11 breaks down the particle placement primes targets into matched and unmatched pairs. On average, unmatched prime-target pairs are separated by eight AS-Units more by comparison to matched pairs in L1-L1 conversations. The same trend can be observed in the L2-L2 conversations where unmatched prime-target pairs were separated by just over two AS-Units more than the matched prime-target pairs. Across data sets, matched prime-target pairs in L2-L2 conversations were separated by an average of eight AS-Units more than matched L1-L1 pairs. Moreover, unmatched prime-target pairs in L1-L1 conversations were separated by an average of two AS-Units less than unmatched prime-target particle placement pairs in the L2-L2 conversations. The VCD package in R studio, (Meyer et al., 2006), was used to produce Figure 4.6 which visualizes the differences between L1-L1 and L2-L2 particle placement prime-target distance (please see section 3.3.2).
158
Figure 4.6: Particle placement priming interaction with prime-target distance in L1-L1 and L2-L2 conversations
The two particle placement prime patterns are shown along the Y-axis in Figure 4.6 above. The two particle placement target patterns are shown along the X- axis. Each bar represents the frequency of particle placement targets within the respective prime-target distance range. What we can see from Figure 4.6 is that for the L1-L1 VP NP PRT primes, the targets were mostly the VP NP PRT pattern, especially when the distance separating prime-target pairs was 12 AS-Units or fewer. It seems that the areas in black, red, green and blue are greater for VP NP PRT targets, relative to VP PRT NP targets following VP NP PRT primes. The VP PRT NP primes, however, along the Y-axis, were followed by more VP NP PRT targets, relative to VP PRT NP targets across all prime-target distances.
For L2-L2 VP NP RT primes, we can see that the area in black, red and green where the prime-target distance is zero, one or less than four AS-Units, respectively, is slightly larger for VP NP PRT targets, relative to VP PRT NP targets. For VP PRT NP primes, however, it seems that the VP NP PRT target pattern was favoured across all prime-target distances.
159 4.5.5.2 Lexical boost
Table 4.12 shows that close to one fifth of the L1-L1 and L2-L2 particle prime-target pairs had a matched main verb lemma. One third of the L1-L1 prime- target pair shared the same particle, while just under one fourth of the L2-L2 prime- target pairs shared the same particle. Finally, slightly less than one third of the L1-L1 prime-target pairs have a shared object, while the L2-L2 prime-target had a matched object only one fourth of the time. A two-sided Fisher’s exact test establishes that there are no significant differences in the proportions of matched and unmatched lemmas or direct objects between L1-L1 and L2-L2 particle placement prime-target pairs (pFisher exact = 0.610) and (pFisher exact = 0.089), respectively. However, the differences between L1-L1 and L2-L2 in terms of shared and unshared particle proportions seem to be significant (pFisher exact = 0.029).
Table 4.12: Particle placement prime-target pairs’ matched and unmatched main verb lemmas, particle, and direct object
Corpus # Matched lemma (%) # Unmatched lemma (%)
L1-L1 44 (19.47%) 182 (80.53%)
L2-L2 32 (17.11%) 155 (82.89%)
Corpus # Matched particle (%) # Unmatched particle (%)
L1-L1 76 (33.63%) 150 (66.37%)
L2-L2 44 (23.53%) 143 (76.47%)
Corpus # Matched D.O. (%) # Unmatched D.O. (%)
L1-L1 66 (29.20%) 160 (70.80%)
L2-L2 40 (21.39%) 147 (82.61%)
Note. L1-L1 = dyadic interaction between participants with English as their L1; L2-L2 = dyadic interaction between participants with English as their L2; # Matched lemma = number of particle placement prime-target pairs sharing the same main verb lemma; # Unmatched lemma = number of particle placement prime-target pairs with different main verb lemmas; # Matched particle = number of particle placement prime-target pairs sharing the same particle; # Unmatched particle = number of particle placement prime-target pairs with different particles; # Matched D.O. = number of particle placement prime-target pairs with the same direct object or its referent; # Unmatched D.O. = number of particle placement prime-target pairs sharing the same direct object or its referent.
160
The proportions of matched and unmatched lemmas, particles and shared objects for particle placement prime-target pairs are shown in Table 4.13. Over two thirds of the L1-L1 prime-target pairs with matched main verb lemmas were of the same verb-particle variant. Four fifths of the L2-L2 prime-target pairs with matched lemmas were of the same verb-particle type. Both the L1-L1 and L2-L2 prime-target pairs with matched particles were of the same verb-particle type two thirds of the time. Finally, the L1-L1 and L2-L2 prime-target pairs with overlapping direct objects were also similar in that they were of the same verb-particle type four fifths of the time.
Table 4.13: Matched and unmatched verb-particle constructions’ prime-target pairs in the case of matched and unmatched main verb lemmas, particles and direct objects
Matched Lemma Unmatched lemma
Corpus # Matched pairs (%) # Unmatched pairs (%) # Matched pairs (%) # Unmatched pairs (%) L1-L1 31 (70.45%) 13 (29.55%) 128 (70.33%) 54 (29.67%) L2-L2 25 (78.13%) 7 (21.87%) 91 (58.71%) 64 (41.29%)
Matched particle Unmatched particle
Corpus # Matched pairs (%) # Unmatched pairs (%) # Matched pairs (%) # Unmatched pairs (%) L1-L1 53 (69.74%) 23 (30.26%) 106 (70.67%) 44 (29.33%) L2-L2 31 (70.45%) 13 (29.55%) 85 (59.44%) 58 (40.56%)
Matched D.O. Unmatched D.O.
Corpus # Matched pairs (%) # Unmatched pairs (%) # Matched pairs (%) # Unmatched pairs (%) L1-L1 53 (80.30%) 13 (19.70%) 106 (66.25%) 54 (34.75%) L2-L2 33 (82.50%) 7 (17.50%) 83 (56.46%) 64 (43.54)
Note. L1-L1 = dyadic interaction between participants with English as their L1; L2-L2 = dyadic interaction between participants with English as their L2; # Matched pairs = number of particle placement primes immediately followed by particle placement targets with the same VP PRT NP or VP NP PRT sequence as in the prime; # Unmatched pairs = number of particle placement primes followed by particle placement targets with a different verb-particle construction to the one in the prime; Matched lemma = particle placement prime-target pairs sharing the same main verb lemma; Unmatched lemma = particle placement prime-target pairs with different main verb lemmas; Matched particle = particle placement prime-target pairs sharing the same particle; Unmatched particle = particle placement prime-target pairs with different particles; Matched D.O. = particle placement prime- target pairs sharing the same direct object or its referent; Unmatched D.O. = particle placement prime-target pairs with different direct objects.
161 4.5.5.3 Priming-speaker interaction
We can see from Table 4.14 that in L1-L1 conversations, almost ninety percent of the primes were followed by targets that were produced by the same speaker. Slightly less than eighty percent of the L2-L2 particle placement prime-target pairs were produced by the same speaker. A two-sided Fisher’s exact test suggests that the differences between the L1-L1 and L2-L2 proportions of prime-target pairs produced by the same or different speakers are statistically significant (pFisher exact = 0.009). Table 4.14: Verb-particle constructions’ prime-target pairs in the case of same and different speaker
Corpus # Same speaker (%) # Different speaker (%)
L1-L1 197 (87.17%) 29 (12.13%)
L2-L2 144 (77.01%) 43 (22.99%)
Note. L1-L1 = dyadic interaction between participants with English as their L1; L2-L2 = dyadic interaction between participants with English as their L2; # Same speaker = number of particle placement prime-target pairs produced by the same speaker; # different speaker = number of particle placement prime-target pairs produced by different speakers.
Table 4.15 shows the proportions of matched and unmatched prime-target pairs in the case where the prime and the target were produced by the same or a different speaker. Just over seventy percent of the L1-L1 prime-target pairs that were produced by the same speaker were matched, i.e. had the same verb-particle
construction in the prime and the target. However, close to three fifths of the L2-L2 prime-target pairs that were produced by the same speaker were of the same verb- particle type.
162
Table 4.15: Matched and unmatched verb-particle constructions’ prime-target pairs when they were produced by the same or different speakers
Same speaker Different speaker
Corpus # Matched pairs (%) # Unmatched pairs (%) # Matched pairs (%) # Unmatched pairs (%) L1-L1 142 (72.08%) 55 (27.92%) 17 (58.62%) 12 (41.38%) L2-L2 89 (61.81%) 55 (38.19%) 27 (62.79%) 16 (37.21%)
Note. L1-L1 = dyadic interaction between participants with English as their L1; L2-L2 = dyadic interaction between participants with English as their L2; # Matched pairs = number of particle placement primes immediately followed by particle placement targets with the same VP PRT NP or VP NP PRT sequence as in the prime; # Unmatched pairs = number of particle placement primes followed by particle placement targets with a different verb-particle construction to the one in the prime; Same speaker = particle placement prime-target pairs produced by the same speaker; Different speaker = particle placement prime-target pairs produced by different speakers.
In the next section, I will discuss the results and highlight the similarities and the differences in the factors that influenced the use of verb-particle types the most in both the L1-L1 and L2-L2 data sets.